Method and device for detecting dysfunction of vehicle embeded computer using digital images

Abstract

The present disclosure concerns a method to train, on a computing device, a machine-learning model adapted to determine a dysfunction of a monitored vehicle electronic control unit (ECU) or vehicle embedded computer. In aspects, the computing device stores, in a memory, historical data from a plurality of ECUs having a dysfunction. The historical data may include usage values over a period of time of at least one ECU resource by applications running on the ECUs. Further, the computing device may process the historical data to obtain two-dimensional training files. In implementations, each usage value may be linked with a specific application in a first dimension and a specific time in a second dimension. Still further, the computing device may train a machine-learning model with the training files.

What is claimed is: 
     
       1. A method comprising:
 storing, in a memory of a computing device, historical data from a plurality of electronic control units (ECUs) having one or more dysfunctions, the historical data including usage values over a period of time of respective ones of a plurality of ECU resources by applications running on a respective one of the plurality of ECUs, the plurality of ECU resources including at least processor usage and memory usage; 
 processing, by the computing device, the historical data to generate three-dimensional training files from digital images having pixels such that coordinates of the pixels in a first dimension are associated with respective ones of the applications, coordinates of the pixels in a second dimension are associated with a runtime for the respective ones of the applications, and each of the pixels has a plurality of channels representing respective ones of a plurality of layers in a third dimension, the plurality of layers including at least a first layer representing the processor usage of each of the applications and a second layer representing the memory usage of each of the applications such that the layers are arranged to correlate the processor usage with the memory usage over the period of time, and intensities of the plurality of channels of the pixels are set to correspond to one of the usage values of the respective ones of the plurality of ECU resources utilized by the applications running on the respective one of the plurality of ECUs at the runtime; and 
 training, by the computing device, a machine-learning model with the three-dimensional training files to determine a dysfunction of a monitored vehicle ECU. 
 
     
     
       2. The method as described in  claim 1 , wherein the historical data include historical data files having different periods of time. 
     
     
       3. The method as described in  claim 1 , wherein the historical data include historical data files having different numbers of applications. 
     
     
       4. The method as described in  claim 1 , further comprising:
 standardizing the three-dimensional training files by scaling up or down at least part of the training files to a predetermined size of the first dimension; or 
 standardizing the three-dimensional training files by scaling up or down at least part of the three-dimensional training files to a predetermined size of the second dimension. 
 
     
     
       5. The method as described in  claim 1 , wherein the three-dimensional training files have a predetermined size in the first dimension. 
     
     
       6. The method as described in  claim 1 , wherein the three-dimensional training files have a predetermined size in the second dimension. 
     
     
       7. The method as described in  claim 1 , wherein the plurality of layers further includes one or more of a third layer representing a network usage of a network adaptor, a fourth layer representing a power consumption, and a fifth layer representing a number of threads. 
     
     
       8. The method as described in  claim 1 , wherein the historical data includes an error message generated by the applications, and wherein the error messages are included in a third layer of the plurality of layers in the three-dimensional training files in order to supervise the training of the machine-learning model. 
     
     
       9. The method of  claim 1 , wherein the intensities of the plurality of channels include at least 8 bits of data such that, over the period of time, the processor usage and the memory usage vary between at least 0 to 255. 
     
     
       10. A method comprising:
 receiving, by a computing device, telemetry data of a monitored vehicle ECU, the telemetry data including usage values over a period of time of respective ones of a plurality of ECU resources by a plurality of applications running on the monitored vehicle ECU and storing the telemetry data in a memory, the plurality of ECU resources including at least processor usage and memory usage; 
 processing, by the computing device, the telemetry data to generate three-dimensional monitoring files from digital images having pixels such that coordinates of the pixels in a first dimension are associated with respective ones of the applications, coordinates of the pixels in a second dimension are associated with a runtime for the respective ones of the applications, and each of the pixels has a plurality of channels representing respective ones of a plurality of layers in a third dimension, the plurality of layers including at least a first layer representing the processor usage of each of the applications and a second layer representing the memory usage of each of the applications such that the layers are arranged to correlate the processor usage with the memory usage over the period of time, and intensities of the plurality of channels of the pixels are set to correspond to one of the usage values of the respective ones of the plurality of ECU resources utilized by the applications running on the monitored vehicle ECU at the runtime; and 
 determining, by the computing device, a dysfunction of the monitored vehicle ECU from the three-dimensional monitoring files input into a machine-learning model trained to determine the dysfunction of the monitored vehicle ECU. 
 
     
     
       11. The method as described in  claim 10 , wherein determining the dysfunction of the monitored vehicle ECU is effective to determine a present or future dysfunction. 
     
     
       12. The method as described in  claim 10 , further comprising:
 standardizing the three-dimensional monitoring files by scaling up or down at least one of the three -dimensional monitoring files to a size of the first dimension or the second dimension of the three-dimensional monitoring files. 
 
     
     
       13. The method as described in  claim 10 , wherein the plurality of layers further includes one or more of a third layer representing a network usage of a network adaptor, a fourth layer representing a power consumption, and a fifth layer representing a number of threads. 
     
     
       14. The method of  claim 10 , wherein the intensities of the plurality of channels include at least 8 bits of data such that, over the period of time, the processor usage and the memory usage vary between at least 0 to 255. 
     
     
       15. A system comprising:
 one or more processors; and 
 a memory coupled to the one or more processors, the memory storing one or more programs configured to be executed by the one or more processors, the one or more programs including instructions that, when executed by the one or more processors, cause the one or more processors to:
 store, in a memory of a computing device, historical data from a plurality of electronic control units (ECUs) having a dysfunction, the historical data including usage values over a period of time of respective ones of a plurality of ECU resources by applications running on a respective one of the plurality of ECUs, the plurality of ECU resources including at least processor usage and memory usage; 
 process, by the computing device, the historical data to generate three-dimensional training files from digital images having pixels such that coordinates of the pixels in a first dimension are associated with respective ones of the applications, coordinates of the pixels in a second dimension are associated with a runtime for the respective ones of the applications, and each of the pixels has a plurality of channels representing respective ones of a plurality of layers in a third dimension, the plurality of layers including at least a first layer representing the processor usage of each of the applications and a second layer representing the memory usage of each of the applications such that the layers are arranged to correlate the processor usage with the memory usage over the period of time, and intensities of the plurality of channels of the pixels are set to correspond to one of the usage values of the respective ones of the plurality of ECU resources utilized by the applications running on the respective one of the plurality of ECUs at the runtime; and 
 
 train, by the computing device, a machine-learning model using the digital images as the three-dimensional training files to determine the dysfunction of a monitored vehicle ECU. 
 
     
     
       16. The system of  claim 15 , wherein the intensities of the plurality of channels include at least 8 bits of data such that, over the period of time, the processor usage and the memory usage vary between at least 0 to 255.